Being CS students interested in Finance, we were naturally drawn to statistical financial modeling and predictive analytics. One of the emerging ways in which people are drawing up their predictions is sentiment analysis. While the concept is sound, we wanted to patch it's inefficiencies and make a more holistic tool that keeps the user more informed of the unforgivingly dynamic landscape of finance.
What it does
The three main forces that we noted moved stock prices were the company financials, how the world felt about the company and its leadership (company sentiment), and external factors such as industry performance, inflation, interest rates, etc. FinTrack not only does sentiment analysis, but our algorithm refines the accuracy of the analysis by taking company financials and other relevant factors into account. We made FinTrack to be an all in one stock analyzer that takes into account all three of these forces and displayed them in a comprehensive manner so the user could have a fast and easy understanding of how the stock is doing right now, and how it will do in the future.
How I built it
Challenges I ran into
• We had a hard time finding good sentiment analyzers online and fetching the news articles due to limited number of API calls.
• Organizing financial data and relating it to stock prices also posed challenges due to missing financial information on websites and the vast economic expertise required to understand them.
• Sentiment and financial analysis for the entire S&P index turned out to be hard because of the vast amount of data we had to deal with.
• Relating external factors like inflation or interest rates to different industries and a specific stock.
Accomplishments that I'm proud of
• Being able to find correlations in our analyser and the market stock prices.
• Patience and relentless persistence displayed by the entire team when we ran into problems with the APIs not giving consistent results and limited number of API calls.
What I learned
• Working with Aylien API to search new articles and parsing them for sentiment analysis.
• Making a well designed and animated frontend.
• Web scraping data for company and industry financials.
• Surviving off of junk food.
What's next for FinTrack
We want to refine our algorithm further by taking more external variables into account and by trial and error, come up with an optimal algorithm that correlates with the stock prices. Our plan is to work with professors in the field of business and finance to develop an all in one sophisticated stock analyzer that quantifies all factors that move stock prices.